After the development of diagnostic tests for hepatitis A virus and hepatitis B virus, an additional agent, which could be experimentally transmitted to chimpanzees (Alter et al, 1978; Hollinger et al, 1978; Tabor et al, 1978), became recognized as the major cause of transfusion-acquired hepatitis. cDNA clones corresponding to the causative non-A non-B (NANB) hepatitis agent, called hepatitis C virus (HCV), were reported in 1989 (Choo et al, 1989). This breakthrough has led to rapid advances in diagnostics, and in our understanding of the epidemiology, pathogenesis and molecular virology of HCV (see Houghton et al, 1994 for review). Evidence of HCV infection is found throughout the world and the prevalence of anti-HCV antibodies ranges from 0.4-2% in most developed countries to more than 14% in Egypt (Hibbs et al, 1993). Besides transmission via blood or blood products, or less frequently by sexual and congenital routes, sporadic cases, not associated with known risk factors, occur and account for more than 40% of HCV cases (Alter et al, 1990; Mast and Alter, 1993). Infections are usually chronic (Alter et al, 1992) and clinical outcomes range from an inapparent carrier state to acute hepatitis, chronic active hepatitis, and cirrhosis which is strongly associated with the development of hepatocellular carcinoma. Although alpha IFN has been shown to be useful for the treatment of some patients with chronic HCV infections (Davis et al, 1989; DiBisceglie et al, 1989) and subunit vaccines show some promise in the chimpanzee model (Choo et al, 1994), future efforts are needed to develop more effective therapies and vaccines. The considerable diversity observed among different HCV isolates (for review, see Bukh et al, 1995), the emergence of genetic variants in chronically infected individuals (Enomoto et al, 1993; Hijikata et al, 1991; Kato et al, 1992; Kato et al, 1993; Kurosaki et al, 1993; Lesniewski et al, 1993; Ogata et al, 1991; Weiner et al, 1991; Weiner et al, 1992), and the lack of protective immunity elicited after HCV infection (Farci et al, 1992; Prince et al, 1992) present major challenges towards these goals.
Molecular biology of HCV
Classification. Based on its genome structure and virion properties, HCV has been classified as a separate genus in the flavivirus family, which includes two other genera: the flaviviruses such as yellow fever virus (YF)! and the animal pestiviruses bovine viral diarrhea virus (BVDV) and classical swine fever virus (CSFV)! (Francki et al, 1991). All members of this family have enveloped virions that contain a positive-strand RNA genome encoding all known virus-specific proteins via translation of a single long open reading frame (ORF; see below).
Structure and physical properties of the virion. Little information is available on the structure and replication of HCV. Studies have been hampered by the lack of a cell culture system able to support efficient virus replication and the typically low titers of infectious virus present in serum. The size of infectious virus, based on filtration experiments, is between 30-80 nm (Bradley et al, 1985; He et al, 1987; Yuasa et al, 1991). HCV particles isolated from pooled human plasma (Takahashi et al, 1992), present in hepatocytes from infected chimpanzees, and produced in cell culture (Shimizu et al, 1994a) have been visualized (tentatively) by electron microscopy. Initial measurements of the buoyant density of infectious material in sucrose yielded a range of values, with the majority present in a low density pool of &lt;1.1 g/ml (Bradley et al, 1991). Subsequent studies have used RT/PCR to detect HCV-specific RNA as an indirect measure of potentially infectious virus present in sera from chronically infected humans or experimentally infected chimpanzees. From these studies, it has become increasingly clear that considerable heterogeneity exists between different clinical samples, and that many factors can affect the behavior of particles containing HCV RNA (Hijikata et al, 1993; Thomssen et al, 1992). Such factors include association with immunoglobulins (Hijikata et al, 1993) or low density lipoprotein (Thomssen et al, 1992; Thornssen et al, 1993). In highly infectious acute phase chimpanzee serum, HCV-specific RNA is usually detected in fractions of low buoyant density (1.03-1.1 g/ml) (Carrick et al, 1992; Hijikata et al, 1993). In other samples, the presence of HCV antibodies and formation of immune complexes correlate with particles of higher density and lower infectivity (Hijikata et al, 1993). Treatment of particles with chloroform, which inactivates infectivity (Bradley et al, 1983; Feinstone et al, 1983), or with nonionic detergents, produces RNA containing particles of higher density (1.17-1.25 g/ml) believed to represent HCV nucleocapsids (Hijikata et al, 1993; Kanto et al, 1994; Miyamoto et al, 1992).
There have been many reports of varying levels of negative-sense HCV-specific RNAs in sera and plasma (see Fong et al, 1991). However, it seems unlikely that such RNAs are essential components of infectious particles since some sera with high infectivity can have low or undetectable levels of negative-strand RNA (Shimizu et al, 1993). The virion protein composition has not been rigorously determined, but putative HCV structural proteins include a basic C protein and two membrane glycoproteins, E1 and E2.
HCV replication. Early events in HCV replication are poorly understood. Cellular receptors for the HCV glycoproteins have not been identified. The association of some HCV particles with beta-lipoprotein and immunoglobulins raises the possibility that these host molecules may modulate virus uptake and tissue tropism. Studies examining HCV replication have been largely restricted to human patients or experimentally inoculated chimpanzees. In the chimpanzee model, HCV RNA is detected in the serum as early as 3 days post-inoculation and persists through the peak of serum alanine aminotransferase (ALT) levels (an indicator of liver damage) (Shimizu et al, 1990). The onset of viremia is followed by the appearance of indirect hallmarks of HCV infection of the liver. These include the appearance of a cytoplasmic antigen (Shimizu et al, 1990) and ultrastructural changes in hepatocytes such as the formation of microtubular aggregates for which HCV previously was referred to as the chloroform-sensitive "tubule forming agent" or "TFA" (reviewed by Bradley, 1990). As shown by the appearance of viral antigens (Blight et al, 1993; Hiramatsu et al, 1992; Krawczynski et al, 1992; Yamada et al, 1993) and the detection of positive and negative sense RNAs (Fong et al, 1991; Gunji et al, 1994; Haruna et al, 1993; Lamas et al, 1992; Nouri Aria et al, 1993; Sherker et al, 1993; Takehara et al, 1992; Tanaka et al, 1993), hepatocytes appear to be a major site of HCV replication, particularly during acute infection (Negro et al, 1992). In later stages of HCV infection the appearance of HCV-specific antibodies, the persistence or resolution of viremia, and the severity of liver disease, vary greatly both in the chimpanzee model and in human patients. Although some liver damage may occur as a direct consequence of HCV infection and cytopathogenicity, the emerging consensus is that host immune responses, in particular virus-specific cytotoxic T lymphocytes, may play a more dominant role in mediating cellular damage (see Rice and Walker, 1995 for review).
It has been speculated that HCV may also replicate in extra hepatic reservoir(s), particularly in chronically infected individuals. In some cases, RT/PCR or in situ hybridization has shown an association of HCV RNA with peripheral blood mononuclear cells including T-cells, B-cells, and monocytes (Blight et al, 1992; Bouffard et al, 1992; Gil et al, 1993; Gunji et al, 1994; Moldvay et al, 1994; Nuovo et al, 1993; Wang et al, 1992; Young et al, 1993; Yun et al, 1993; Zignego et al, 1992). Such tissue tropism could be relevant to the establishment of chronic infections and might also play a role in the association between HCV infection and certain immunological abnormalities such as mixed cryoglobulinemia (reviewed by Ferri et al, 1993), glomerulonephritis, and rare non-Hodgkin's B-lymplhomas (Ferri et al, 1993; Kagawa et al, 1993). However, the detection of circulating negative strand RNA in serum, the difficulty in obtaining truly strand-specific RT/PCR (Gunji et al, 1994), and the low numbers of apparently infected cells have made it difficult to obtain unambiguous evidence for replication in these tissues in vivo.
Although a cell culture system capable of efficient HCV replication has not been developed, some progress has been made. Consistent with the in vivo observations mentioned above, in vitro HCV infection and short term replication have been reported for chimpanzee and human hepatocytes (Carloni et al, 1993; lacovacci et al, 1993; Lanford et al, 1994), a human hepatoma line (Huh7; Yoo et al, 1995, see below), peripheral blood leukocytes (Muller et al, 1993), a human B-cell line expressing EBV antigens (Bertolini et al, 1993), a mouse retrovirus-infected human T-cell line (Molt4-Ma; Shimizu et al, 1992), an HTLV-1 transformed human T-cell line (MT-2; Kato et al, 1995), and fibroblasts derived from human foreskin (Zibert et al, 1995). Thus far, only a small fraction of these cells appear infected. In vitro infectivity of different HCV inocula using a permissive subclone of the Molt4-Ma T-cell line correlates well with their in vivo infectivity in the chimpanzee model (Shimizu et al, 1993). This cell line has also been used to begin examining HCV binding and the possible emergence of neutralization escape mutants during chronic infection (Shimizu et al, 1994b).
Genome structure. Full-length or nearly full-length genome sequences of numerous HCV isolates have been reported (see Lin et al, 1994; Okainoto et al, 1994; Sakamoto et al, 1994 and citations therein). Given the considerable genetic divergence among isolates, it is clear that several major HCV genotypes are distributed throughout the world (see below). Those of greatest importance in the U.S. are genotype 1, subtypes 1a and 1b. HCV genome RNAs are about 9.4 kilobases in length. The 5' NTR is 341-344 bases and is the most conserved RNA sequence element in the HCV genorne. The length of the long ORF varies slightly among isolates, encoding polyproteins of 3010, 3011 or 3033 amino acids. The reported 3' NTR structures show considerable diversity both in composition and length (28-42 bases), and appear to terminate with poly (U) (for examples, see Chen et al, 1992; Okamoto et al, 1991; Tokita et al, 1994) except in one case (HCV-1, type 1a) which appears contain a 3' terminal poly (A) tract (Han et al, 1991).
Translation and proteolytic processing. Several studies have used cell-free translation and transient expression in cell culture to examine the role of the 5' NTR in translation initiation (Fukushi et al, 1994; Tsukiyaina-Kohara et al, 1992; Wang et al, 1993; Yoo et al, 1992). This highly conserved sequence contains multiple short AUG-initiated ORFs and shows significant homology with the 5' NTR region of pestiviruses (Bukh et al, 1992; Han et al, 1991). A series of stem-loop structures have been proposed on the basis of computer modeling and sensitivity to digestion by different ribonucleases (Brown et al, 1992; Tsukiyama-Kohara et al, 1992). Although still controversial (see Wang et al, 1993; Yoo et al, 1992), the results from several groups indicate that this element functions as an internal ribosome entry site (IRES) allowing efficient translation initiation at the first AUG of the long ORF (Fukushi et al, 1994; Tsukiyama-Kohara et al, 1992; Wang et al, 1993). Some of the predicted features of the HCV and pestivirus IRES elements are similar to one another (Brown et al, 1992). It has been proposed that the 5' terminal hairpin structure and the short ORFs may function to downregulate translation (Yoo et al, 1992). The ability of this element to function as an IRES suggests that HCV genorne RNAs may lack a 5' cap structure.
The organization and processing of the HCV polyprotein appears to be most similar to that of the pestiviruses. At least 10 polypeptides have been identified and the order of these cleavage products in the polyprotein is NH.sub.2 -C-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B-COOH. Proteolytic processing is mediated by host signal peptidase and two HCV-encoded proteinases, the NS2-3 autoproteinase and the NS3-4A serine proteinase. C is a basic protein believed to be the viral core or capsid protein; E1 and E2 are putative virion envelope glycoproteins; p7 is a hydrophobic protein of unknown function that is inefficiently cleaved from the E2 glycoprotein (Lin et al, 1994; Mizushima et al, 1994; Selby et al, 1994), and NS2-NS5B are likely nonstructural (NS) proteins which function in viral RNA replication complexes. In particular, besides its N-terminal serine proteinase domain, NS3 contains motifs characteristic of RNA helicases and has been shown to possess an RNA-stimulated NTPase activity (Suzich et al, 1993); NS5B contains the GDD motif characteristic of the RNA-dependent RNA polymerases of positive-strand RNA viruses.
Virion assembly and release. This process has not been examined directly, but the lack of complex glycans, the ER localization of expressed HCV glycoproteins (Dubuisson et al, 1994; Ralston et al, 1993) and the absence of these proteins on the cell surface (Dubuisson et al, 1994; Spaete et al, 1992) suggest that initial virion morphogenesis may occur by budding into intracellular vesicles. Thus far, efficient particle formation and release has not been observed in transient expression assays, suggesting that essential viral or host factors are missing or blocked. HCV virion formation and release may be inefficient, with a substantial fraction of the virus remaining cell-associated, as found for the pestiviruses. A recent study indicates that extracellular HCV particles partially purified from human plasma do contain complex N-linked glycans, although these carbohydrate moieties were not shown to be specifically associated with E1 or E2 (Sato et al, 1993). Complex glycans associated with glycoproteins on released virions would suggest transit through the trans Golgi and movement of virions through the host secretory pathway. If this suggestion is correct, intracellular sequestration of HCV glycoproteins and virion formation might then play a role in the establishment of chronic infections by minimizing immune surveillance and preventing lysis of virus-infected cells via antibody and complement.
Genetic variability. As for all positive-strand RNA viruses, the RNA-dependent RNA polymerase of HCV (NS5B) is believed to lack a 3'-5' exonuclease proof reading activity for removal of misincorporated bases. Replication is therefore error-prone leading to a "quasispecies" virus population consisting of a large number of variants (Martell et al, 1992; Martell et al, 1994). This variability is apparent at multiple levels. First, in a chronically infected individual changes in the virus population occur over time (Ogata et al, 1991; Okamoto et al, 1992) and these changes may have important consequences for disease. A particularly interesting example is the N-terminal 30 residues of the E2 glycoprotein which exhibits a much higher degree of variability than the rest of the polyprotein (for examples, see Higashi et al, 1993; Hijikata et al, 1991; Weiner et al, 1991). There is accumulating evidence that this hypervariable region, perhaps analogous to the V3 domain of HIV-1 gp120, may be under immune selection by circulating antiviral antibodies (Kato et al, 1993; Taniguchi et al, 1993; Weiner et al, 1992). In this model, antibodies directed against this portion of E2 may contribute to virus neutralization and thus drive the selection of variants with substitutions which escape neutralization. This plasticity suggests that a specific amino acid sequence in the E2 hypervariable region is not essential for other functions of the protein such as virion attachment, penetration, or assembly. Genetic variability may also contribute to the spectrum of different responses observed after treatment of chronically infected patients with alpha IFN. Diminished serum ALT levels and improved liver histology, which is sometimes correlated with a decrease in the level of circulating HCV RNA, is seen in only .about.40% of those treated (Greiser-Wilke et al, 1991). After treatment, approximately 70% of the responders relapse. In some cases, after a transient loss of circulating viral RNA, renewed viremia is observed even during the course of treatment. While this might suggest the existence or generation of IFN-resistant HCV genotypes or variants, further work is needed to determine the relative contributions of virus genotype and host-specific differences in immune responsiveness. Finally, sequence comparisons of different HCV isolates around the world have uncovered enormous genetic diversity (reviewed in Ref. Bukh et al, 1995). Because biologically relevant serological assays such as cross-neutralization tests are lacking, HCV types (designated by numbers), subtypes (designated by letters), and isolates are currently being grouped on the basis of nucleotide or amino acid sequence similarity. Amino acid sequence similarity between the most divergent genotypes can be as little as .about.50%, depending upon the protein being compared. This diversity is likely to have important biological implications, particularly for diagnostics, vaccine design, and therapy. As mentioned earlier, genotypes la and lb are most common in the U.S. (see Bukh et al, 1995 for a discussion of genotype prevalence and distribution). Recently, in Yoo et al (1995) T7 transcripts from various derivatives of an HCV-1 cDNA clone were tested for their ability to replicate by transfection of the human hepatoma cell line, Huh7. Possible HCV replication was assessed by strand-specific RT/PCR (using 5' NTR primers) and metabolic labeling of HCV-specific RNAs with 3H-uridine. Transcripts terminating with either poly (A) or poly (U), were positive by these assays but those with a deletion of the 5' terminal 144 bases were not. In some cultures, HCV-specific RNA could be detected in the culture media and could be used to reinfect fresh Huh7 cells. While these claims cannot be directly refuted, it seems likely that the authors are not actually detecting authentic HCV replication. For instance, the authors' positive control was productive transfection of Huh7 cells with RNA extracted from 1 ml of high HCV titer chimpanzee plasma. This extracted sample would contain a maximum of 107 potentially infectious full-length HCV RNA molecules. Under optimum transfection conditions (other than microinjection), &gt;105 RNA molecules of virion RNA (at least for poliovirus, Sindbis virus, or YF) are typically required to initiate a single infectious event. This suggests that in the HCV-1 experiment fewer than 100 cells would be productively transfected. At 16 days post-transfection, both positive- and negative-strand RNAs were readily detected after 8 hours of metabolic labeling. The detection of negative-strand RNA by this method (both for transfected virion RNA and transcript RNA) suggests that HCV is capable of both efficient replication and spread, and that the level of HCV RNA synthesis is similar to that which would be expected for a more robust flavivirus, such as YF (at the peak of a high multiplicity infection). However, despite numerous attempts, the authors were unable to detect HCV antigens in these cells using a variety of antisera or full-length positive- or negative-strands by Northern analysis (which is much more sensitive than metabolic labeling with 3H-uridine) (J. Han, personal communication). To say the least, these results are perplexing and not easily reconciled with authentic HCV replication. Finally, the critical experiment, demonstrating that RNA or virus derived from the HCV-1 clone is infectious in the chimpanzee model, has not been reported (despite the initial presentation of this work at a meeting more than two years ago). Work in other RNA virus systems has shown that specific terminal sequences can be critical for the generation of functional, replication competent RNAs (reviewed in Boyer and Haenni, 1994). Such sequences are believed to be involved in initiation of negative- and positive-strand RNA synthesis. In some cases, a few additional bases, or even longer non-viral sequences, are tolerated at the 5' and 3' termini; these sequences are typically lost or selected against during authentic viral replication. For other RNA viruses, extra bases, particularly at the 5' terminus, are deleterious (Boyer and Haenni, 1994). In contrast, except in a few cases, transcripts lacking authentic terminal sequences are non-functional (Boyer and Haenni, 1994). For instance, deletion of the 3' terminal secondary structure or conserved sequence elements in the 3' NTR of flavivirus genome RNA is lethal for YF (P. J. Bredenbeek and C.M.R., unpublished) or TBE (C. Mandl, personal communication) RNA replication. Given the importance of these sequence elements for other viruses, it is clear that a more rigorous determination of the HCV terminal sequences needed to be made.